Genesis-Embodied-AI/genesis-world

Simulation platform for general-purpose robotics & embodied AI learning.

View on GitHub ↗Jump to charts ↓Open shareable report

Summary Information

Updated 49 minutes ago
Added to GitGenius on February 27th, 2026
Created on October 31st, 2023
Open Issues & Pull Requests: 133 (+0)
Number of forks: 2,833
Total Stargazers: 29,721 (+2)
Total Subscribers: 224 (+0)

Issue Activity (beta)

Open issues: 105
New in 7 days: 13
Closed in 7 days: 5
Avg open age: 226 days
Stale 30+ days: 84
Stale 90+ days: 71

Recent activity

Opened in 7 days: 9
Closed in 7 days: 4
Comments in 7 days: 7
Events in 7 days: 21

Top labels

  • bug (465)
  • enhancement (207)
  • P2 (74)
  • P0 (57)
  • P1 (56)
  • question (24)
  • documentation (23)
  • good first issue (2)

Repository Insights (GitGenius)

Median issue/PR response: 0.0 hours
Mean response time: 4.3 days
90th percentile: 39.2 hours
Tracked items: 1,075

Most active contributors

Sign in to see contributor activity.

Detailed Description

Genesis World is a simulation platform designed for physical AI development and embodied AI research, combining a unified multi-physics engine, a photo-realistic renderer called Nyx, and a cross-platform compiler named Quadrants behind a Python-based simulation interface. The platform is engineered to scale from single laptop kernels to datacenter-grade GPUs while maintaining accessibility for research code integration. Originally launched as an academic project in December 2024, Genesis World has transitioned to official development support from Genesis AI.

The platform's architecture consists of four integrated layers. The simulation interface provides the user-facing API with asset parsing capabilities for multiple formats including URDF, MJCF, OBJ, GLB, and USD, alongside entity accessors, controllers, sensors, parallel and heterogeneous environment support, and a built-in GUI. The physics layer implements a unified multi-physics engine that integrates rigid body dynamics, finite element method (FEM), material point method (MPM), particle-based dynamics including both position-based dynamics (PBD) and smoothed particle hydrodynamics (SPH), the uipc library, an explicit coupler, and SAP, all operating within a shared scene and state. The rendering layer exposes three distinct rendering paths as camera sensors: Nyx (an in-house renderer optimized for robotics), Luisa (a domain-specific language ray tracer), and Pyrender (a rasterizer). The compiler layer uses Quadrants to lower Python kernel code to multiple backends including CUDA, AMD ROCm, Apple Metal, Vulkan, x86, and ARM64 architectures while preserving Genesis's autodifferentiation, GPU graph capabilities, and fastcache machinery.

The repository demonstrates extensive physics simulation capabilities through its catalogue of examples spanning rigid body dynamics, deformable objects, fluids, and coupled multi-physics scenarios. Physics examples include rigid body manipulation with robotic arms, collision dynamics, finite element constraints, material point method simulations, particle-based liquid dynamics, cloth simulation, smoke dynamics, and implicit contact handling. The rendering examples showcase camera control systems, entity following, animated viewpoints, and advanced Nyx features including physically-based materials, various light types, 3D Gaussian splatting, object picking, and multi-camera multi-environment setups.

genesis-world
by
Genesis-Embodied-AIGenesis-Embodied-AI/genesis-world

Repository Details

Fetching additional details & charts...